Index Morganagus, Easter report
Eric Lease Morgan
eric_morgan at ncsu.edu
Sat Mar 29 08:19:31 EST 1997
:
Index Morganagus, Easter report
This report briefly evaluates the usefulness Index
Morganagus, the index of library-related electronic
serials at:
http://sunsite.berkeley.edu/~emorgan/morganagus/
This report can be found at:
http://sunsite.berkeley.edu/~emorgan/morganagus/
survey-results/report-easter.html
Service extended
When the Index was introduced in late December, a time
limit evaluation period was placed on the service,
Easter. It is almost Easter and it has been decided to
extend the evauation period until Labor Day (September
2) of this year. At that time the usefulness of the
service will be re-evaluated and a decision will be
made whether or not to continue.
Lots of statistics
Computer services are great for generating lots of
data. The Index is no exception. HTTP server logs have
been kept and analysed, as well as Harvest queries, and
survey results.
Survey results
In an attempt to get feedback, a survey was created.
Most of the people answering the survey were academic
librarians. Most people felt the Index was "useful" or
"somewhat useful". To most people the Index returned
the expected results and would like to see the Index
continue. Pie charts illustrating the survey results
are available online.
The public comments and anonymous comments were very
interesting to read. Most of them liked the service but
were wondering whether or not the layout of the search
results could be improved.
Desite the fact that the Index has been searched at
least 3,000 times, only 100 surveys have been recived.
Why? How is a person expected to improve a service if
there is no feedback?
Types of searches
To date, the Index has services 3,123 queries broken
down into the following types:
Phrase 766
Truncation 138
Logical AND 2005
Logical OR 96
Compound 67
Field 34
Single term 1479
Total searches 3123
Based on this data, and a look at the searches
themselves, it is evident people are using single terms
or non-quoted phrases as their primary means of
searching the Index. This is unfortunate becuase
Harvest provides quite robust searching syntax
including field searching, truncation, compound
searches, and truncation.
Put another way, despite the fact that most of the
people using the Index are librarians and expected to
be expert searchers, the librarians are not taking full
advantage of the Index's capabilities. Since the Index
creates logical AND queries by default, and this stated
clearly in the instructions, many queries returned
inexact results. Example queries include:
* public AND libraries AND management AND finance
* flexible AND scheduling
* executive AND board AND room AND table AND designs
* electronic AND citations
* digital and library and finance
Automated indexing
We are happy to report that maintenance of the Index
has bee rather painless after its original
implementation and construction of cron-based shell
scripts. The Index refreshes itself every two weeks
without incident. Its almost "automagtic!" The creation
of the FileMaker Pro database has also been a boon
making simplifying the process of updating the index.
Future directions
Things are changing on the 'Net and the version of
Harvest being used for the Index must be updated to
interpret HTML 3.2 and HTTP 1.1. Another addition is
the ability to interpret PDF files for indexing.
The Index is not perfect. One problem that was brought
to attention was poor indexing of the Newsletter on
Serial Pricing Issues. This conical site for this
serial was being indexinged by Harvest but returning
incorrect results. Why this was happening was not
discovered. Instead, a different archive of the serial
is being used and search results are now correct.
Another opportunity it to join the Index with other
Harvest index around the 'Net to essencially create the
Internet equivelant of Academic Index. This is still
the primary goal of the entire project, but first some
of the interface bugs need to be worked out.
--
Eric Lease Morgan
NCSU Libraries
http://www.lib.ncsu.edu/staff/morgan/
More information about the Web4lib
mailing list